Extract tables from a PDF
Drop a PDF below: every table in it comes back as rows and columns you can download as an Excel file or a CSV, or copy into a spreadsheet. It is free, needs no account, and your file is deleted as soon as the tables are returned. It works on PDFs that contain text, not on scans.
No PDF at hand? Try a sample:
How it works
What it handles
Limits
- Scanned PDFs and photos are not read: they have no text layer. Use a tool with OCR for those.
- Tables are recognised by text aligned in columns. A table drawn without clear columns can be missed, and text laid out in columns can be taken for a table.
- Files up to 2 MB, and 10 files per hour per visitor. For more, use the API below.
Other ways to extract tables from a PDF
| Excel (Microsoft 365, Windows) | Data, Get Data, From File, From PDF imports tables through Power Query. Good if you have that version; amounts in European formats often need cleaning afterwards. |
|---|---|
| Tabula | Free desktop application for text PDFs. You select the area of each table by hand, which is precise but slow for many files. |
| Camelot (Python) | Open source library for developers. Flexible, but you write and maintain the code. |
| General PDF to Excel converters | They turn every page into cells, text included. You then delete what is not a table. |
Automate it: API and AI agents
This tool runs PDF Tables (holon-labs/pdf-tables), an open source agent on Holon. Your code, or an AI agent such as Claude, can call it for many documents: 0.004 EUR per successful run, and a document where no table is found is not billed.
From Claude, once connected over MCP
> Upload statement.pdf to Holon, extract its tables, and total the amounts by month.
API reference, code samples and measured results · Connect Claude to Holon
Questions
Is it really free?
Yes: up to 10 files per hour, without an account. Developers and AI agents who need more call the same agent through the API, at 0.004 EUR per document where tables are found. A document without a table costs nothing.
Do you keep my file?
No. The file is stored only while the tables are extracted, then deleted, at most 15 minutes after the upload. The tables are not kept either. See the privacy policy.
Does it work on scanned PDFs?
No. It reads the text inside the PDF. A scan is an image of text, so the tool answers that no table was found. You can check: if you cannot select the text in your PDF viewer, the PDF is a scan.
What does the Excel file contain?
One sheet per table, the header row first. Numbers and amounts are real numbers, so sums and filters work; the currency of an amount column is in its header.
Can I extract tables from many PDFs automatically?
Yes. The same extraction is available as an API and to AI agents such as Claude over MCP, billed per document. See the developer page of the agent.
How your file is handled: privacy policy.